Blog posts

2022

Motivating Interpretability - A Researcher and Engineering Perspective

2 minute read

Published:

Deep neural networks (DNNs) are known to be almost fully opaque compared to traditional algorithms, even to top engineers and scientists who use DNNs. Many reasons have been proposed for why understanding the decision making process of these models is important, such as human curiosity, scientific discovery, bias detection, or algorithm safety measures and auditing. Furthermore, a model which is interpretable may exhibit more fairness, reliability, and trust to the general public 1. Several methods to interpret DNNs in computer vision have been proposed to varying degrees of success. For example, many recent SOTA saliency map visualization methods have been shown to be on par with some random baselines. Similar issues with other interpretability methods have been criticized on social media, pointing out that the reliance on these methods in mission critical situations will likely do more harm than good.

2020

Understanding Mutual Information

9 minute read

Published:

For this post I hope to accomplish a few different things. I will review a brilliant document put together by Erik G. Learned-Miller. I found this document when attempting to better understand the concept of ‘Mutual Information’, and it has been by far the most influential document in my understanding of entropy and mutual information. It is a short document, only 3 pages, with the purpose of being an introduction to entropy and mutual information for discrete random variables. Erik does something that I think so many teachers miss when introducing students to new concepts, which is using real-world, easy to understand examples to aid with the formulas. I hope I can add some intuition behind what entropy, joint entropy, and mutual information actually represent, as well as review some simple and more complex examples that I am currently working on in my research.